In this chapter, we analyse the practices and experiences of people with diabetes who develop, use and share open-source, non-regulated “recipes” for automating insulin delivery with personal digital health technology. The algorithmic systems are known as Open-Source Artificial Pancreas Systems and the algorithm-enabled activity that these people engage in is often referred to as “looping”. Through empirical accounts from the rich and complex practice of using open-source algorithms in diabetes self-management we explore how this concept of looping may hold the potential to critically explore and discuss more general issues related to human-algorithms relations in digital health. We suggest three ways in which looping holds general insights about the potential for more generous human-algorithm relations. First, looping as an active delegation of control given an existing burden of self-care contingent on the acquisition of new skills; second, looping as a collective and recursive engagement with (material) politics of care and data; and third, looping as the ability to opt-out—partly or totally—of toxic intimate entanglement with algorithmic technologies and of extractivist algorithmic assemblages.